After a long absence from posting due to heavy workload, I wanted to answer this request/question by RJOh regarding the predictability performance of my system with a concrete and official mechanism. I refer to this old discussion here where RJOh wanted to monitor my prediction ability.
https://www.lotterypost.com/thread/240293
I'll not post results as they are posted automatically to this system as they are generated. I wanted to point out just some observations, such as the Euromillions game I inspect right now.
This is a real prediction performance over the last 8 draws on Euromillions 5/50 (as of 9 September 2016) when picking 18 numbers (each prediction was performed before the draw obviously):
1 time a 1-hit
4 times a 3-hit
3 times a 4-hit
Given that a 3+ hit has a natural chance of 24.12% or once in 4.14 draws, I think nothing needs to be explained here. This particular instance shows the underlying concept of reduced randomness is there, the system trapped it and you get this outcome. Similar results can be observed in other games analyzed too, like a hot hit streak of 7 consecutive draws (12 March 2016 to 2 April 2016) of the Canada Lotto 6/49 where it returned (when picking 18 numbers):
1x 3+0
1x 3+1
1x 4+0
2x 4+1
2x 5+1
Just to show to outstanding performance here, the chance to have one 5+1 hit is once in 149 draws (0.66%) and a 4+1 hit once in 21.5 draws (4.6%). This again clearly demonstrates a reduced randomness effect taking place and trapped by the system. Such observations are happening every so often.
Quite good results pop up even if we utilize fewer numbers, quite better compared to natural probability.